RealSR-R1: Reinforcement Learning for Real-World Image Super-Resolution with Vision-Language Chain-of-Thought

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Qiao, Junbo, Cai, Miaomiao, Li, Wei, Huang, Xudong, Hu, Jie, Chen, Xinghao, Lin, Shaohui, Xiong, Hongkai
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917400665915392
author Qiao, Junbo
Cai, Miaomiao
Li, Wei
Huang, Xudong
Hu, Jie
Chen, Xinghao
Lin, Shaohui
Xiong, Hongkai
author_facet Qiao, Junbo
Cai, Miaomiao
Li, Wei
Huang, Xudong
Hu, Jie
Chen, Xinghao
Lin, Shaohui
Xiong, Hongkai
contents Real-World Image Super-Resolution is one of the most challenging task in image restoration. However, existing methods struggle with an accurate understanding of degraded image content, leading to reconstructed results that are both low-fidelity and unnatural. We present RealSR-R1 in this work, which empowers the RealSR models with understanding and reasoning capabilities. Inspired by the success of Chain of Thought (CoT) in large language models (LLMs), we simulate the human process of handling degraded images and propose the VLCoT framework, which integrates vision and language reasoning. The framework aims to precisely restore image details by progressively generating more comprehensive text and higher-resolution images. To overcome the challenge of traditional supervised learning CoT failing to generalize to real-world scenarios, we introduce, for the first time, Group Relative Policy Optimization (GRPO) into the Real-World Image Super-Resolution task. We propose VLCoT-GRPO as a solution, which designs four reward functions: (1) Format reward, used to standardize the CoT process; (2) Degradation reward, to incentivize accurate degradation estimation; (3) Understanding reward, to ensure the accuracy of the generated content; and (4) Generation reward, where we propose using a visual expert model to evaluate the quality of generated images, encouraging the model to generate more realistic images. Extensive experiments demonstrate that our proposed RealSR-R1 can generate realistic details and accurately understand image content, particularly in semantically rich scenes or images with severe degradation.
format Preprint
id arxiv_https___arxiv_org_abs_2506_16796
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RealSR-R1: Reinforcement Learning for Real-World Image Super-Resolution with Vision-Language Chain-of-Thought
Qiao, Junbo
Cai, Miaomiao
Li, Wei
Huang, Xudong
Hu, Jie
Chen, Xinghao
Lin, Shaohui
Xiong, Hongkai
Computer Vision and Pattern Recognition
Real-World Image Super-Resolution is one of the most challenging task in image restoration. However, existing methods struggle with an accurate understanding of degraded image content, leading to reconstructed results that are both low-fidelity and unnatural. We present RealSR-R1 in this work, which empowers the RealSR models with understanding and reasoning capabilities. Inspired by the success of Chain of Thought (CoT) in large language models (LLMs), we simulate the human process of handling degraded images and propose the VLCoT framework, which integrates vision and language reasoning. The framework aims to precisely restore image details by progressively generating more comprehensive text and higher-resolution images. To overcome the challenge of traditional supervised learning CoT failing to generalize to real-world scenarios, we introduce, for the first time, Group Relative Policy Optimization (GRPO) into the Real-World Image Super-Resolution task. We propose VLCoT-GRPO as a solution, which designs four reward functions: (1) Format reward, used to standardize the CoT process; (2) Degradation reward, to incentivize accurate degradation estimation; (3) Understanding reward, to ensure the accuracy of the generated content; and (4) Generation reward, where we propose using a visual expert model to evaluate the quality of generated images, encouraging the model to generate more realistic images. Extensive experiments demonstrate that our proposed RealSR-R1 can generate realistic details and accurately understand image content, particularly in semantically rich scenes or images with severe degradation.
title RealSR-R1: Reinforcement Learning for Real-World Image Super-Resolution with Vision-Language Chain-of-Thought
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.16796